Appetite 58 (2012) 11–18

Contents lists available at SciVerse ScienceDirect

Appetite

journal homepage: www.elsevier .com/locate /appet

Research report

The role of energetic value in dynamic brain response adaptation duringrepeated food image viewing

Claudia V. Lietti b, Micah M. Murray a,b,c, Julie Hudry d, Johannes le Coutre d,e, Ulrike Toepel a,b,⇑a The Functional Electrical Neuroimaging Laboratory, Department of Clinical Neurosciences, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerlandb The Functional Electrical Neuroimaging Laboratory, Department of Radiology, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerlandc EEG Core, Center for Biomedical Imaging of Lausanne and Geneva, Switzerlandd Nestlé Research Center, Vers-chez-les-Blanc, Lausanne, Switzerlande The University of Tokyo, Organization for Interdisciplinary Research Projects, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan

a r t i c l e i n f o

Article history:Received 20 May 2011Received in revised form 15 September2011Accepted 24 September 2011Available online 8 October 2011

Keywords:EEGERPFood perceptionVisionRepetitionAdaptation

0195-6663/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.appet.2011.09.016

⇑ Corresponding author.E-mail address: Ulrike.Toepel@chuv.ch (U. Toepel)

a b s t r a c t

The repeated presentation of simple objects as well as biologically salient objects can cause the adapta-tion of behavioral and neural responses during the visual categorization of these objects. Mechanisms ofresponse adaptation during repeated food viewing are of particular interest for better understanding foodintake beyond energetic needs. Here, we measured visual evoked potentials (VEPs) and conducted neuralsource estimations to initial and repeated presentations of high-energy and low-energy foods as well asnon-food images. The results of our study show that the behavioral and neural responses to food andfood-related objects are not uniformly affected by repetition. While the repetition of images displayinglow-energy foods and non-food modulated VEPs as well as their underlying neural sources and increasedbehavioral categorization accuracy, the responses to high-energy images remained largely invariantbetween initial and repeated encounters. Brain mechanisms when viewing images of high-energy foodsthus appear less susceptible to repetition effects than responses to low-energy and non-food images. Thisfinding is likely related to the superior reward value of high-energy foods and might be one reason why inparticular high-energetic foods are indulged although potentially leading to detrimental healthconsequences.

� 2011 Elsevier Ltd. All rights reserved.

Introduction

The prevalence of obesity is increasing throughout the worldpopulation, causing long-term health problems such as diabetesor cardiac disorders. Obesity is linked to a loss in the ability to ad-just food intake for maintaining the energetic balance of the body.Deviant eating behavior leading to obesity can, however, not beseen as a pure problem of food intake. There is increasing evidencethat human food intake is not only controlled by brain areas in-volved in homeostatic control, but also by cortical and subcorticalareas involved in the reward and cognitive aspects of hedonic feed-ing (Gibson, Carnell, Ochner, & Geliebter, 2010). Moreover, nutri-tion intake behavior is strongly based on pre-ingestion choicesthat can be driven by visual food appearance, including influencesfrom marketing strategies such as prices (Knutson, Rick, Wimmer,Prelec, & Loewenstein, 2007) as well as repeated product exposure(Krajbich, Armel, & Rangel, 2010).

ll rights reserved.

.

Our previous research has demonstrated that food images arereadily differentiated according to their energetic content,resulting in power and topography modulations of the electric fieldwhen high-fat vs. low-fat foods are viewed (Toepel, Knebel, Hudry,le Coutre, & Murray, 2009). As low-level visual features were con-trolled for, this differentiation is likely related to the varying foodreward properties of high- and low-energetic foods as well deci-sion mechanisms related to their pre-ingestion valuation. Whetherthe repeated exposure to these food classes alters the brain pro-cesses underlying the differentiation of foods have, however, notyet been investigated.

So far, the adaptation of behavioral and neural responses to re-peated object exposure has mainly been investigated by employingstimuli like familiar and unfamiliar objects and faces. Typically,behavioral effects manifest through facilitation of reaction timesand accuracy rates for the discrimination and categorization ofrepeatedly as opposed to initially presented objects. Hemodynamicneuroimaging studies often report activation changes in prefrontaland ventral temporal cortex during repeated object exposure – aneffect typically referred to as repetition priming (Dolan et al.,1997; George et al., 1999; Grill-Spector, Henson, & Martin, 2006;Henson, Shallice, & Dolan, 2000 for review). Electrophysiological

12 C.V. Lietti et al. / Appetite 58 (2012) 11–18

studies often show early amplitude modulations of the visualevoked potentials (VEP) around �150–200 ms post-stimulus onset(George, Jemel, Fiori, & Renault, 1997; Guillaume et al., 2009;Henson, 2003; Henson, Rylands, Ross, Vuilleumier, & Rugg, 2004;Schendan & Kutas, 2003; Soldan, Mangels, & Cooper, 2006; Taci-kowski, Jednoróg, Marchewka, & Nowicka, 2011). However, laterrepetition-induced VEP modulations are frequently observed aswell around 250 and 400 ms (Henson, 2003; Schweinberger, Pfütze,& Sommer, 1995; Schweinberger, Pickering, Burton, & Kaufmann,2002). Therein, the alterations in behavioral and neural responsepatterns are most often explained in terms of implicit memoryprocesses (e.g. De Lucia et al., 2010; Henson, 2003; Murray, Camen,Spierer, & Clarke, 2008; Schacter, Dobbins, & Schnyer, 2004).

What previous studies leave open, however, is the question ofwhether repetition-induced neural and behavioral adaptationmechanisms are also effective when the viewed objects are poten-tially relevant for bodily energy homeostasis, i.e. food items. More-over, when comparing response adaptations for food classes thatdiffer in energetic content, varying perceptual and mnemonicmechanisms might be indexed as high- and low-energetic foodsinherently differ with respect to their reward and hedonic proper-ties. High-energy foods have, for example, been shown to be morepleasant even in normal-weighted adults (Finlayson, King, & Blun-dell, 2007) with such preferences already shaped during childhood(Birch, 1999). As lower food response adaptation or habituation,respectively, is associated with greater food intake (Temple, Gia-comelli, Roemmich, & Epstein, 2007; Wisniewski, Epstein, & Caggi-ula, 1992), more invariant representations of high-energetic foodsacross repetitions (as compared with low-energy food items andnon-food artifacts) might contribute to why especially high-ener-getic foods are indulged despite of potential long-term detrimentalconsequences (like obesity, and in turn, increased risk for cardiovas-cular disorders). For this purpose, raw data from our previous study(Toepel et al., 2009) were processed and analyzed with a focus oncomparing VEPs and behavioral responses to the initial and re-peated presentations of images depicting high- and low-energeticfoods (in terms of fat content) as well as non-food kitchen utensils.

Material and methods

Participants

Twenty-one remunerated volunteers partook in the study (11women, 10 men, aged 19–33 years [mean ± s.e.m. = 24.62 ± 1.01 -years]). All of these 21 individuals were from our original study(Toepel et al., 2009). 17 participants were right-handed, and fourambidextrous according to the Edinburgh Handedness Inventory(Oldfield, 1971). Data from one additional subject were excludedfrom the present analyses due to insufficient signal quality. Allsubjects had a BMI within the normal range (mean ± -s.e.m = 21.95 ± 0.53 kg/m2) and no current or prior neurologicalor psychiatric illness or self-reported eating disorders. They allhad normal or corrected-to-normal vision. Each EEG recording ses-sion started between 13h00 and 14h00 to control for circadianmodulations of hunger. Moreover, all participants were instructedand also self-reported to have eaten lunch before the recordingsession. Participants provided written, informed consent to theprocedures, which were approved by the Ethics Committee of thefaculty of Biology and Medicine of the University of Lausanneand the Vaudois University Hospital Center.

Stimuli and procedure

The present analyses focused on the initial and first repetition ofimages presented in our previous experiment (Toepel et al., 2009).

That is, the effects of repeated object exposure were studied on100 photographs of foods and 50 non-food images (Fig. 1). The foodimages were subdivided into equi-sized high- and low-energy clas-ses by means of their fat content (cf. Toepel et al., 2009) and are here-after referred to as HiFat and LoFat categories, though we wouldemphasize that fat and energy content are highly correlated. Thetemporal lag between the initial and repeated presentation of an im-age was �8–10 min due to image randomization within blocks.

All pictures were adapted in luminance at an individual imagelevel (Knebel, Toepel, Hudry, le Coutre, & Murray, 2008; see Toepelet al., 2009 for further information). However, due to the predom-inant goal of our previous analyses, the spatial frequency distribu-tion between the food categories and the non-food image categorycould not be fully adapted (Fig. 1, right panel).

The pictures measured 300 � 300 pixels, which corresponded to�6� visual angle on the computer monitor and had been photo-graphed using an identical background and top-view angle. Eachimage was presented for 500 ms in the center of a 2100 CRT monitorthat participants viewed within an electrically shielded and soundattenuated booth. Immediately after the picture presentation, aquestion mark was presented on the screen as a request to decidevia button-press if the preceding image had been a food or a non-food item. The question mark remained on the screen until a re-sponse was made, allowing participants to self-pace the experi-ment. In order to minimize eye movements, a crosshair wasshown in the screen center whenever no image or question markwas present. The interval between crosshair and image onset var-ied randomly between 250 and 750 ms. Stimulus presentation andresponse recordings were controlled by E-Prime (Psychology Soft-ware Tools Inc., Pittsburgh, USA; www.pstnet.com/eprime).

EEG data acquisition and pre-processing

Continuous electroencephalogram (EEG) was acquired at512 Hz through a 160-channel Biosemi Active Two system (Bio-semi, Amsterdam, Netherlands) referenced to the common modesense/driven right leg (CMS-DRL) ground. All data pre-processingand analyses were performed with the CarTool software (http://si-tes.google.com/site/fbmlab/cartool). VEP epochs from 98 ms pre-to 488 ms post-stimulus onset (i.e. 50 data points before and 250data points after stimulus onset) were first separately averagedfor each presentation (initial and repeated), image category (HiFat,LoFat and NoFood), and each participant. An ±80 lV artifact rejec-tion criterion was applied to the dataset and EEG epochs contain-ing eye blinks or other noise transients were removed throughtrial-by-trial visual inspection. Before performing the group aver-aging for each presentation and condition, data from artifact-con-taminated electrodes were interpolated (Perrin, Pernier, Bertrand,Giard, & Echallier, 1987). The dataset was also baseline-correctedusing the pre-stimulus period, band-pass filtered (0.1–40 Hz) andrecalculated against the average reference. The number of VEPepochs (mean and s.e.m. given) per category entering the statisticalanalyses did not differ significantly between presentation phases(HiFat initial vs. repeated: 48.62 [±0.38] vs. 48.86 [±0.32];p = 0.51; LoFat initial vs. repeated: 48.38 [±0.38] vs. 48.38[±0.52]; p = 1.00; NoFood initial vs. repeated: 47.38 [±0.39] vs.46.86 [±0.48]; p = 0.17).

EEG analyses

General analysis strategyThe electrophysiological analyses applied used both local and

global measures of the electric field at the scalp. These so-calledelectrical neuroimaging analyses allow differentiating effectscaused by modulations in the VEP strength or amplitude, respec-tively, from alterations in the VEP topography considering data

Fig. 1. (left panel) Exemplar stimuli from each image category; i.e. HiFat and LoFat foods, and NoFood kitchen utensils. (right panel) Histograms for each image categoryillustrating the similarity in spatial frequency distribution between HiFat and LoFat food images.

C.V. Lietti et al. / Appetite 58 (2012) 11–18 13

from all electrode sensors (here: 160) in a concurrent manner. Onlyfundamental information on the analyses is given here (see Michelet al., 2004 and Murray, Brunet, & Michel, 2008 for further details).For all local and global VEP analyses, only effects with p-val-ues 6 0.05 were considered statistically significant (with Green-house-Geisser correction applied when the sphericity assumptionwas violated) and temporal correlation was corrected throughthe application of a 15 contiguous data-point criterion (�30 ms;Guthrie & Buchwald, 1991).

As pure effects of image category in our design could be con-founded by low-level visual feature differences between image cat-egories (i.e. spatial frequencies; see right panel of Fig. 1) we willhenceforth emphasize effects of presentation and interactions be-tween image category and presentation phase.

VEP waveform analysesThe first analysis step included a 3 � 2 sample-wise ANOVA

across all electrodes involving the factors presentation (initial, re-peated) and category (HiFat, LoFat and NoFood). The output ofthe analysis is an intensity plot indicating time, electrode locationand significant p-values at each data point. These analyses on localVEP modulations give a visual impression of effects within thedataset and provide a link between electrical neuroimaging andmore traditional VEP analysis approaches. We would note, how-ever, that our conclusions are based on reference-independent glo-bal measures of the electric field.

A global topographic cluster analysis was performed to identifythe sequence of VEP topographies within and across categories andpresentation phases (Murray, Brunet, et al., 2008; Murray, Camen,et al., 2008). Topographic VEP modulations indicate differences inthe brain’s underlying generators (Lehmann, 1987) independentfrom strength modulations of the electric field. The optimal numberof topographic maps was determined using a modified Krzanowski-Lai criterion, which is a measure of the dispersion across clusters(see Murray, Brunet, et al., 2008; Murray, Camen, et al., 2008 for for-mulae). When map topographies were found to descriptively differby presentation and/or category over a certain time period, the ob-served differences at the group level were statistically validated bycomparing the map cluster appearance with the individuals’ VEPsfrom each presentation phase and category over a given time peri-od. This analysis is referred to as ‘‘fitting’’. During the fitting proce-dure, each time point of each single subject VEP was labeled inaccordance to the group-average map topography with which itbest correlated spatially (Murray, Brunet, et al., 2008; Murray, Ca-men, et al., 2008). The dependent measure is the global explainedvariance (GEV) in percent, indicating how well a certain map iden-tified in the group-averaged data explains the VEP responses from a

given individual participant and for each presentation phase (initialor repeated) or image category viewed (HiFat, LoFat and NoFood).Repeated measures ANOVAs with the factors presentation phase,category and map were performed to validate presentation- andcategory-related modulations in map topography presence. Post-hoc paired t-tests were conducted when appropriate.

Importantly, the fitting procedure does not test whether theelectric field configurations of the best-correlating maps them-selves statistically differ. This aspect was addressed by performinga time point-wise topographic ANOVA or T-ANOVA, respectively,that provides a direct index of global spatial VEP dissimilarity inde-pendent of electric field strength (Murray, Brunet, et al., 2008;Murray, Camen, et al., 2008). In detail, the T-ANOVA tested in5000 randomizations per data sampling point whether the VEPtopography to each image category and presentation statisticallydiffers from the mean VEP topography across presentations andcategories (Koenig, Melie-García, Stein, Strik, & Lehmann, 2008;Wirth et al., 2008).

In addition to the topographic analyses, modulations in the glo-bal VEP strength were assessed using global field power (GFP; Leh-mann & Skrandies, 1980) at each data point. GFP is calculated asthe square root of the mean of the squared value recorded at eachelectrode in the 160-channel montage (vs. the average reference)and represents the spatial standard deviation of the electric fieldat the scalp. It yields larger values for stronger electric fields. Sta-tistical differences in GFP were assessed in a 3 � 2 ANOVA includ-ing the factors presentation (initial, repeated) and category (HiFat,LoFat and NoFood) and post-hoc paired t-tests when appropriate.

Estimation of neural sources underlying VEPsWe estimated the intracranial sources generating the VEPs for

each image category and presentation phase using the local auto-regressive average (LAURA) distributed linear inverse solution(Grave de Peralta, Gonzalez Andino, Lantz, Michel, & Landis,2001; Grave de Peralta Menendez, Murray, Michel, Martuzzi, &Gonzalez Andino, 2004). The version of LAURA used here employsa realistic head model with 3005 nodes arranged within the graymatter of the Montreal Neurological Institute’s (MNI) averagebrain, and was generated with the Spherical Model with Anatom-ical Constraints (SMAC; Spinelli, Andino, Lantz, Seeck, & Michel,2000). As an output, LAURA provides current density value (inlA/mm3) at each node. Prior fundamental and clinical researchhave documented and discussed in detail the spatial accuracy ofthis inverse solution, which are on the order of the grid size ofthe solution points (here �6 � 6 � 6 mm, Gonzalez Andino, Michel,Thut, Landis, & Grave de Peralta, 2005; Gonzalez Andino, Murray,Foxe, & de Peralta Menendez, 2005; Grave de Peralta Menendez

14 C.V. Lietti et al. / Appetite 58 (2012) 11–18

et al., 2004; Michel et al., 2004; Michel, Seeck, & Murray, 2004).The time period for which intracranial sources were estimatedand statistically compared between presentation phases (here:125–164 ms post-image onset) was defined by the topographiccluster analysis. Statistics on the source estimations was per-formed by first averaging the VEP data over the 125–164 ms inter-val to generate a single data point for each participant to increasethe signal-to-noise ratio. The inverse solution (21 participants � 3image categories � 2 presentations) was then estimated for each ofthe 3005 nodes. An ANOVA was performed with the within-subjectfactors of image category (HiFat, LoFat and NoFood) and presenta-tion (initial vs. repeated) at each source node. Post-hoc paired t-tests between initial and repeated image viewing were conductedfor each image category separately (using the ANOVA output asinclusive mask). The results of these analyses were rendered onthe MNI brain with the corresponding coordinates of Talairachand Tournoux (1988) and similarly the locus of the Brodmann areaof the maximal t-value indicated. Only effects with p-values 6 0.05and present in at least 15 contiguous nodes were considered signif-icant (cf. Toepel et al., 2009). This spatial criterion was determinedusing the AlphaSim program (available at http://afni.nimh.nih.-gov). 10,000 Monte Carlo permutations were performed usingthe nodes of our lead field matrix and revealed a false positiveprobability of <0.005 for observing a cluster of at least 15 nodes.No additional correction for multiple comparisons was applied.

Results

Behavioral results

Table 1 provides a summary of accuracy rates and reactiontimes in the food vs. non-food discrimination task. A 3 � 2 ANOVAon the accuracy rates with the factors of presentation and categoryrevealed main effects of category (F(2,40) = 9.69; p 6 0.01) and ofpresentation (F(1,20) = 9.52; p 6 0.01) as well as an interaction be-tween these factors (F(2,40) = 9.69; p 6 0.01).

Post-hoc paired t-tests revealed significant improvements inaccuracy between initial vs. repeated presentations for LoFat foods(t(20) = �2.29; p 6 0.05) and NoFood (t(20) = �4.08; p 6 0.01), butnot for HiFat foods. Moreover, accuracy during initial presentationswas found to be higher for HiFat foods than for NoFood(t(20) = 3.42; p 6 0.01), and higher for LoFat foods than for NoFood(t(20) = 3.58; p 6 0.01).

An 3 � 2 ANOVA on reaction times with the factors of presenta-tion and category showed only a main effect of presentation(F(1,20) = 22.36; p 6 0.01). Participants’ reaction times decreasedreliably during the repeated viewing of all three image categoriesas compared to the initial presentation (HiFat: t(20) = 2.87;p 6 0.05; LoFat: t(20) = 4.72; p 6 0.05; NoFood: t(20) = 5.40;p 6 0.05). We would remind the reader that RT speed was notemphasized in the instructions to the participants or in the para-digm itself.

Results of the local and global VEP analyses

Figure 2a shows the results of the electrode- and time-point-wise ANOVA on all electrodes and Fig. 2b the group-averaged

Table 1Response accuracy (in percent) and reaction times (in ms) in the behavioral object catego

Initial presentation

RT in ms (±s.e.m.) Accuracy in % (±s.e.m

HiFat 443.55 (29.45) 97.28 (1.60)LoFat 452.71 (28.95) 96.50 (1.56)NoFood 454.49 (30.46) 92.20 (1.96)

VEP waveforms at three exemplar electrodes. The ANOVA evincedan effect of presentation across a wide range of electrodes starting�130 ms post-image onset and a sustained effect of category from�190 ms. No significant interactions including neighboring elec-trodes were observed by the time-point wise ANOVA at the singlewaveform level.

As these local and reference-dependent VEP analyses cannot re-veal the bases of the repetition-induced differences (i.e. whetherthey arise from alterations in strength or topography of the electricfield), we conducted a common topographic cluster analysis on thegroup-averaged VEPs to all image categories (HiFat, LoFat and No-Food) and presentation phases (initial, repeated). This analysisidentified periods of eight stable electric field topographic clusters(maps) that explained 98.78% of the variance in the collectivegroup-average dataset. The sequence of topographic clusters wassimilar across object categories and presentation phases over themajority of the post-stimulus period. Differing map clusters wereobserved from 125 to 164 ms (see Fig. 3a), i.e. over the period thatis in temporal agreement with the effect of presentation observedat single electrode level (see Fig. 2a).

Over the time interval from 125 to 164 ms, two map topogra-phies (map A and B in the left panel of Fig. 3a) were first identifiedat the VEP group-average level. These maps were mainly character-ized by a fronto-medial VEP maximum with negative amplitude inmap A, and a more left-lateralized frontal negativity in map B (alsoapparent in the difference between map A and B). The occurrence ofthe maps as a function of image category and presentation phasewas statistically validated in the VEPs from individual subjects bythe fitting procedure using a 3 � 2 � 2 ANOVA. The analysis re-vealed a main effect of presentation (F(1,20) = 5.82; p 6 0.05) andan interaction of category � presentation �map (F(2,40) = 5.10;p 6 0.05). Such an interaction indicates that the extent to which agiven template map accounted for the initial vs. repeated presenta-tion of a particular image type itself significantly differed. This canmore simply be understood as indicating that different patterns ofresponse adaptation as expressed in the VEP topography were tak-ing place for different image types. Post-hoc paired t-tests focusingon the impact of repetition revealed changes in the pattern of thesemap topographies when LoFat foods and NoFood images wereviewed, but not when HiFat food images were viewed. In particular,during LoFat food viewing, map A (framed in black) was better rep-resentative of subjects’ responses during repeated than initial pre-sentations (t(20) = 2.15; p 6 0.05). By contrast, when NoFood imageswere viewed, map B (framed in gray) reliably better accounted forsubjects’ responses during repeated than initial presentations(t(20) = 2.87; p 6 0.01). That is, the VEP topographies elicited byviewing LoFat and NoFood objects were modulated by whetherthe images were seen for the first or second time. However, re-sponses to HiFat foods did not result in altered VEP topographies.

The topographic map cluster analysis, although well suited tovalidate the contribution of single subjects to the group-averagetopographic VEP responses, does not test in how far the (descrip-tively) varying map topographies over a certain time period them-selves differ. In order to test for global VEP dissimilarity, we thusperformed a time-point wise topographic ANOVA over the post-stimulus epoch (T-ANOVA; see ‘‘Methods’’ section). The T-ANOVArevealed an effect of presentation between 125 and 193 ms and

rization task during initial and repeated image encounters.

Repeated presentation

.) RT in ms (±s.e.m.) Accuracy in % (±s.e.m.)

389.41 (24.13) 100.00 (0)372.44 (24.04) 100.00 (0)373.82 (26.32) 100.00 (0)

Fig. 2. (a) Results of the electrode- and time-point-wise 2 � 3 ANOVA. Spatially and temporally distributed effects of presentation began �130 ms, while effects of imagecategory started at �190 ms post-image onset. (b) Top panel: Exemplar electrodes showing VEP waveforms jointly for all conditions with the latencies of the main effect ofpresentation indicated by red bars. I = initial presentation, R = repeated presentation. Lower panels: Separate electrode displays for each category (HiFat, LoFat and NoFood)illustrating the VEP waveforms during initial (I) and repeated (R) image presentations.

C.V. Lietti et al. / Appetite 58 (2012) 11–18 15

from 375 ms to the end of the computation period. A temporallysustained effect of image category, on the other hand, only became

evident later, i.e. from 181 ms. In accordance with the previouslyobtained repetition-induced differences by image category at the

Fig. 3. (a) The topographic clustering on the collective group-averaged VEPs revealed a time interval of stable but varying VEP topographies (i.e. over the 125–164 ms post-stimulus interval). The bar graphs show the results of the fitting procedure for the interval, expressing the global explained variance (GEV in percent) a given map topographyyielded a higher spatial correlation with the single-subject responses than an alternative one (±s.e.m. indicated). Bars framed in black illustrate the variance explained by maptopography A (also framed in black), and bars framed in gray illustrate the variance explained by map topography B (also framed in gray). Asterisks above the bar graphsindicate significant values in the post-hoc paired t-tests between presentation phases within one image category; I = initial presentation, R = repeated presentation. Below thetopographic map displays, the difference topography comparing map A–B is visualized. (b) Differences in neural source estimations underlying the observed VEPs as obtainedby an ANOVA including the responses to all image categories and both presentations, and separate paired t-tests for each image category comparing the neural sourceactivation during initial vs. repeated image presentations.

16 C.V. Lietti et al. / Appetite 58 (2012) 11–18

topographic map cluster level, separate T-ANOVAs were conductedfor each category. These yielded reliable differences in global VEPtopography between initial and repeated presentations of LoFatfoods (144–187 ms) and NoFood images (142–158 and 175–189 ms), but not when HiFat foods were viewed.

As all the analyses on topographic VEP modulations areindependent of electric field strength, we further analyzed glo-bal field power (GFP) to test whether VEP response differencesare also reflected in modulations of response strength. A time-point wise ANOVA on GFP with the factors image categoryand presentation over the peri-stimulus epoch evinced only ef-fects of category (i.e. between 218–291 and 330–408 ms) but

no changes in global response strength as a function of presen-tation phase.

Results of the neural source estimations on VEP generators

The upper panel of Fig. 3b illustrates the interaction betweenimage category (HiFat, LoFat and NoFood) and presentation (initialvs. repeated) on the neural source activity underlying the VEPsobserved on the scalp-surface over the time interval from 125 to164 ms. Significant interactions between image category and pre-sentation were evident in the prefrontal and the middle temporalcortex, mostly restricted to the left hemisphere. Source nodes

C.V. Lietti et al. / Appetite 58 (2012) 11–18 17

showing significant interactions were rendered on the MNI averagebrain for visualization.

Separate post-hoc paired t-tests for each image category com-paring the neural source strength between presentation phases(lower panel of Fig. 3b) showed that activity in the ventral prefron-tal cortex of the left hemisphere was lower when LoFat images wereviewed for the second as opposed to the first time (Max: �54 8 13;BA4; t(20) = 3.26). In contrast, the viewing of NoFood images in-duced elevated activation in the middle temporal cortex (Max:�63 �27 0; BA21; t(20) = �3.68) of the left hemisphere when itemswere encountered for the second as opposed to first time. When Hi-Fat images were viewed, no reliable source estimation differenceswere observed between initial and repeated presentations.

Discussion

The results of our study reveal that visual processing of food andnon-food objects is not uniformly affected by repetition. In terms ofbehavioral adaptation in the object categorization task, participantsperformed at near-ceiling level during initial image presentationsand at ceiling level during repeated exposure. Reliable increasesin accuracy following repetition were limited to objects depictinglow-energy foods and non-food kitchen utensils. In contrast, nosuch improvement was observed during the repeated exposure tohigh-energy food images. During initial presentation, high-andlow-energy foods were categorized with similar accuracy, indicat-ing that the lack in response adaptation between the initial and re-peated viewing of high-energy foods could not solely be attributedto a general recognition advantage for high-energy foods. More-over, reaction times decreased between initial and repeated view-ing of all three image categories, an effect frequently reported forrepetition-induced behavioral modulations (Henson, 2003). Yet,we would like to emphasize that participants were cued to respondonly after image onset (i.e. to avoid confounds of VEP modulationsand motor-evoked brain responses), so that the observed decreasesin reaction time have to be interpreted with caution.

The time-point wise analyses on VEP waveforms (Fig. 2a) indi-cated as a first result that repetition-induced modulations precedeeffects induced by image categorization. In this respect, our resultsare consistent with findings on repetition priming for faces, i.e. an-other class of biologically highly salient objects. Such studies eithershowed direct modulations of face-sensitive VEP components likethe N170 (Eimer, Kiss, & Nicholas, 2010; Guillaume et al., 2009;Itier & Taylor, 2004) or even repetition-induced modulations pre-ceding face-selective categorization effects (George et al., 1997;Michel, Seeck & Murray, 2004; Seeck et al., 1997).

Moreover, our data revealed differential effects depending onwhether images of high-or low-energy foods or non-food objectswere repeatedly presented. While wide-spread repetition-inducedVEP modulations were observed for low-energy food and non-foodimages, the VEPs to high-energy food images were less altered dur-ing presentation phases. A topographic cluster analysis incorporat-ing data points from all 160 electrode sensors across the full post-stimulus period then identified the time interval between 125 and164 ms as bearing stable VEP topographic maps that differed, how-ever, depending on whether low-energy foods or non-food itemswere seen for the first or the second time. No topographic modula-tion was, on the other hand, observed when high-energy foodimages were repeatedly viewed.

These repetition-induced topographic VEP differences immedi-ately precede what we previously identified as the latency whenimages of high-energy and low-energy foods were incidentally dis-criminated from each other (i.e. at �165 ms; Toepel et al., 2009). Inour previous study, neural source estimations revealed the differ-ential engagement of a network mostly encompassing prefrontal

and temporo-occipital areas in the visual perception of high- vs.low-energy foods, i.e. brain regions that are implicated into objectcategorization, the assessment of food reward and decision-mak-ing (e.g. Killgore et al., 2003).

Our current study points to alterations in neural networkrecruitment imposed by repetition priming. Yet, the implicit mem-ory processes that priming effects are usually associated with dif-fered by the type of image category that was repeatedlyencountered. Repeated viewing of non-food items led to enhancedresponses in middle temporal cortex, a brain region proximate toinferior temporal areas often reported in imaging studies on facepriming (see Grill-Spector et al., 2006; Henson, 2003 for reviews).On the other hand, the repeated encounter of images depictinglow-energy foods resulted in response suppression in ventral pre-frontal areas. In contrast, we did neither observe repetitionenhancement nor suppression when images of high-energy foodswere repeatedly viewed although high- and low-energy foodimages were closely matched in terms of low-level visual features(cf. Fig. 1), and both image categories comprised of commonly con-sumed and easily identifiable exemplars. Object repetition is oftenassociated with response suppression, although response enhance-ment has previously been observed as well, i.e. during face percep-tion. According to extant accounts, response enhancement isthought to indicate processes during item repetition that werenot performed during initial exposure reflecting, e.g., a stabiliza-tion of mnemonic traces due to increasing stimulus familiarity ordiscrete item recognition within a category (Fiebach, Gruber, &Supp, 2005; Henson, 2003; Henson et al., 2000; Wiggs & Martin,1998). On the other hand, response suppression supposedly re-flects similar brain processes during initial and repeated exposure,e.g., an efficient reactivation or ‘‘sharpening’’ of representations offamiliar objects. In keeping with these accounts, (low-energy) foodand non-food objects likely undergo a differential treatment duringrepeated encounter. While the re-recognition of (low-energy) foodobjects seem to involve the efficient exploitation of a mnemonictrace established already during initial viewing, repeating non-food items likely leads to further mnemonic consolidation based,e.g., on the recognition of specific object-determining features.Notably, the finding of ventral prefrontal (for low-energy foods)vs. middle temporal (for non-food items) repetition modulationsfurther indicates that food as a primary reinforcer and reward im-pacts implicit memory processes differently than non-food objects,but also other biologically salient stimuli like faces (Grill-Spectoret al., 2006; Henson, 2003).

However, when bringing together our behavioral, global VEPand neural source findings, the representations of high-energyfoods appear to be more invariant between initial and repeatedviewing than those of low-energy foods. High-energy foods areknown to have a strong impact on homeostatic body balance, toproduce stronger hedonic drives than low-energy foods and areassociated with higher reward (for overviews see Almiron-Roig &Drewnowski, 2003; Rolls, 2009) and greater motivational salience(Frank et al., 2010). That is, these properties likely render high-energetic foods mnemonically potent and stable, and, in turn, lesssusceptible to repetition-induced modulations than low-energeticfoods, even when only perceived visually, i.e. evaluated for poten-tial consumption. The more invariant representation of high-ener-getic foods throughout first and repeated encounters might be acontributing (but surely not exclusive) reason why especiallyhigh-energetic foods are eaten beyond energetic body homeostasisalthough bearing potentially detrimental long-term consequencesfor body weight and health. Another possibility, which we cannotfully discount, is that our analysis methods lacked sufficient sensi-tivity to detect differences between initial and repeated presenta-tions of high-energetic foods. We consider this unlikely, as VEPs tothis class of images were based on similar numbers of EEG epochs

18 C.V. Lietti et al. / Appetite 58 (2012) 11–18

as those to other classes of images. Additionally, because our elec-trical neuroimaging analyses entailed independent tests of topog-raphy and global field power, we had internal replication ofpositive effects for low-energetic and non-food images and replica-tion of negative effects for high-energetic images. When coupledwith additional analyses linked to the topographic cluster analysisas well as distributed source estimations (both of which also failedto reveal significant effects for high-energetic food images), arecurring pattern emerges.

Notwithstanding, our study only tapped into one aspect ofmemory for food and our analysis included only instances of firstand second food item encounters. Thus, we cannot exclude thatfurther visual encounters of high-energy foods would not lead tobehavioral and neural repetition modulations as found for low-en-ergy food items. Yet, also slower habituation to food has been asso-ciated with greater food intake (Wisniewski et al., 1992; Templeet al., 2007). Along these lines, weight gain is often related to gen-erally altered responses in brain areas implicated in reward assess-ment, so that more extended investigations on memory processesfor food in overweight women and men would be beneficial to bet-ter understand how to interact with hedonic food intake beyondhomeostatic needs, i.e. in order to develop cognitive-behavioralweight management strategies.

References

Almiron-Roig, E., & Drewnowski, A. (2003). Hunger, thirst, and energy intakesfollowing consumption of caloric beverages. Physiology and Behavior, 79,767–773.

Birch, L. L. (1999). Development of food preferences. Annual Review of Nutrition, 19,41–62.

De Lucia, M., Cocchi, L., Martuzzi, R., Meuli, R. A., Clarke, S., & Murray, M. M. (2010).Perceptual and semantic contributions to repetition priming of environmentalsounds. Cerebral Cortex, 20, 1676–1684.

Dolan, R. J., Fink, G. R., Rolls, E., Booth, M., Holmes, A., & Frackowiak, R. S. J. (1997).How the brain learns to see objects and faces in an impoverished context.Nature, 389, 596–599.

Eimer, M., Kiss, M., & Nicholas, S. (2010). Response profile of the face-sensitive N170component. A rapid adaptation study. Cerebral Cortex, 20, 2442–2452.

Fiebach, C. J., Gruber, T., & Supp, G. G. (2005). Neuronal mechanisms of repetitionpriming in occipitotemporal cortex. Spatiotemporal evidence from functionalmagnetic resonance imaging and electroencephalography. Journal ofNeuroscience, 25, 3414–3422.

Finlayson, G., King, N., & Blundell, J. E. (2007). Is it possible to dissociate ‘liking’ and‘wanting’ for foods in humans? A novel experimental procedure. Physiology andBehavior, 90, 36–42.

Frank, S., Laharnar, N., Kullmann, S., Veit, R., Canova, C., Hegner, Y. L., Fritsche, A., &Preissl, H. (2010). Processing of food pictures. Influence of hunger, gender andcalorie content. Brain Research, 1350, 159–166.

George, N., Raymond, J. D., Fink, G. R., Baylis, G. C., Russell, C., & Driver, J. (1999).Contrast polarity and face recognition in the human fusiform gyrus. NatureNeuroscience, 2, 574–580.

George, N., Jemel, B., Fiori, N., & Renault, B. (1997). Face and shape repetition effectsin humans. A spatio-temporal ERP study. NeuroReport, 8, 1417–1423.

Gibson, C. D., Carnell, S., Ochner, C. N., & Geliebter, A. (2010). Neuroimaging, gutpeptides and obesity. Novel studies of the neurobiology of appetite. Journal ofNeuroendocrinology, 8, 833–845.

Gonzalez Andino, S. L., Murray, M. M., Foxe, J. J., & de Peralta Menendez, R. G. (2005).How single-trial electrical neuroimaging contributes to multisensory research.Experimental Brain Research, 166, 298–304.

Gonzalez Andino, S. L., Michel, C. M., Thut, G., Landis, T., & Grave de Peralta, R.(2005). Prediction of response speed by anticipatory high-frequency (gammaband) oscillations in the human brain. Human Brain Mapping, 24, 50–58.

Grave de Peralta, R., Gonzalez Andino, S. L., Lantz, G., Michel, C. M., & Landis, T.(2001). Noninvasive localization of electromagnetic epileptic activity: 1. Methoddescriptions and simulations. Brain Topography, 14, 131–137.

Grave de Peralta Menendez, R., Murray, M. M., Michel, C. M., Martuzzi, R., &Gonzalez Andino, S. L. (2004). Electrical neuroimaging based on biophysicalconstraints. Neuroimage, 21, 527–539.

Grill-Spector, K., Henson, R., & Martin, A. (2006). Repetition and the brain. Neuralmodels of stimulus-specific effects. Trends in Cognitive Sciences, 10, 14–23.

Guillaume, C., Guillery-Girard, B., Chaby, L., Lebreton, K., Hugueville, L., Eustache, F.,& Fiori, N. (2009). The time course of repetition effects for familiar faces andobjects. An ERP study. Brain Research, 1248, 149–161.

Guthrie, D., & Buchwald, J. S. (1991). Significance testing of difference potentials.Psychophysiology, 28, 240–244.

Henson, R. N., Shallice, T., & Dolan, R. (2000). Neuroimaging evidence for dissociableforms of repetition priming. Science, 287, 1269–1272.

Henson, R. N. (2003). Neuroimaging studies of priming. Progress in Neurobiology, 70,53–81.

Henson, R. N., Rylands, A., Ross, E., Vuilleumier, P., & Rugg, M. D. (2004). The effect ofrepetition lag on electrophysiological and hemodynamic correlates of visualobject priming. Neuroimage, 21, 1674–1689.

Itier, R. J., & Taylor, M. J. (2004). Effects of repetition learning on upright, invertedand contrast-reversed face processing using ERPs. Neuroimage, 21, 1518–1532.

Killgore, W. D., Young, A. D., Femia, L. A., Bogorodzki, P., Rogowska, J., & Yurgelun-Todd, D. A. (2003). Cortical and limbic activation during viewing of high- versuslow-calorie foods. Neuroimage, 19, 1381–1394.

Knebel, J.-F., Toepel, U., Hudry, J., le Coutre, J., & Murray, M. M. (2008). Methods forgenerating controlled image sets in cognitive neuroscience research. BrainTopography, 20, 284–290.

Knutson, B., Rick, S., Wimmer, G. E., Prelec, D., & Loewenstein, G. (2007). Neuralpredictors of purchases. Neuron, 53, 147–156.

Koenig, T., Melie-García, L., Stein, M., Strik, W., & Lehmann, C. (2008). Establishingcorrelations of scalp field maps with other experimental variables usingcovariance analysis and resampling methods. Clinical Neurophysiology, 119,1262–1270.

Krajbich, I., Armel, C., & Rangel, A. (2010). Visual fixations and the computation andcomparison of value in simple choice. Nature Neuroscience, 13, 1292–1298.

Lehmann, D. (1987). Principles of spatial analysis. In A. S. Gevins & A. Reymond(Eds.). Handbook of electroencephalography and clinical neurophysiology. Methodsof analysis of brain electrical and magnetic signals (vol. 1, pp. 309–354).Amsterdam: Elsevier.

Lehmann, D., & Skrandies, W. (1980). Reference-free identification of components ofcheckerboard-evoked multichannel potential fields. Electroencephalography andClinical Neurophysiology, 48, 609–621.

Michel, C. M., Murray, M. M., Lantz, G., Gonzalez, S., Spinelli, L., & Grave de Peralta, R.(2004). EEG source imaging. Clinical Neurophysiology, 115, 2195–2222.

Michel, C. M., Seeck, M., & Murray, M. M. (2004). The speed of visual cognition.Clinical Neurophysiology, 57, 617–627.

Murray, M. M., Brunet, D., & Michel, C. M. (2008). Topographic ERP analyses. A step-by-step tutorial review. Brain Topography, 20, 249–264.

Murray, M. M., Camen, C., Spierer, L., & Clarke, S. (2008). Plasticity inrepresentations of environmental sounds revealed by electrical neuroimaging.Neuroimage, 39, 847–856.

Oldfield, R. C. (1971). The assessment and analysis of handedness. The EdinburghInventory. Neuropsychologia, 9, 97–113.

Perrin, F., Pernier, J., Bertrand, O., Giard, M. H., & Echallier, J. F. (1987). Mapping ofscalp potentials by surface spline interpolation. Electroencephalography andClinical Neurophysiology, 66, 75–81.

Rolls, B. J. (2009). The relationship between dietary energy density and energyintake. Physiology and Behavior, 97, 609–615.

Schacter, D. L., Dobbins, I. G., & Schnyer, D. M. (2004). Specificity of priming. Acognitive neuroscience perspective. Nature Reviews Neuroscience, 5, 853–862.

Schendan, H. E., & Kutas, M. (2003). Time course of processes and representationssupporting visual object identification and memory. Journal of CognitiveNeuroscience, 15, 111–135.

Schweinberger, S. R., Pfütze, E. M., & Sommer, W. (1995). Repetition priming andassociative priming of face recognition. Evidence from event-related potentials.Journal of Experimental Psychology, Learning, Memory and Cognition, 21, 722–736.

Schweinberger, S. R., Pickering, E. C., Burton, A. M., & Kaufmann, J. M. (2002). Humanbrain potential correlates of repetition priming in face and name recognition.Neuropsychologia, 40, 2057–2073.

Seeck, M., Michel, C. M., Mainwaring, N., Cosgrove, R., Blume, H., Ives, J., Landis, T., &Schomer, D. L. (1997). Evidence for rapid face recognition from human scalp andintracranial electrodes. NeuroReport, 8, 2749–2754.

Soldan, A., Mangels, J. A., & Cooper, L. A. (2006). Evaluating models of object-decision priming. Evidence from event-related potential repetition effects.Journal of Experimental Psychology: Learning, Memory, and Cognition, 32,230–248.

Spinelli, L., Andino, S. G., Lantz, G., Seeck, M., & Michel, C. M. (2000). Electromagneticinverse solutions in anatomically constrained spherical head models. BrainTopography, 13, 115–125.

Tacikowski, P., Jednoróg, K., Marchewka, A., & Nowicka, A. (2011). How multiplerepetitions influence the processing of self-, famous and unknown names andfaces. An ERP study. International Journal of Psychophysiology, 79, 219–230.

Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas of the human brain. 3-dimensional proportional system. An approach to cerebral imaging. New York:Thieme Medical Publishers.

Temple, J. L., Giacomelli, A. M., Roemmich, J. N., & Epstein, L. H. (2007). Overweightchildren habituate slower than non-overweight children to food. Physiology andBehavior, 91, 250–254.

Toepel, U., Knebel, J.-F., Hudry, J., le Coutre, J., & Murray, M. M. (2009). The braintracks the energetic value in food images. Neuroimage, 44, 967–974.

Wiggs, C. L., & Martin, A. (1998). Properties and mechanisms of perceptual priming.Current Opinions in Neurobiology, 8, 227–233.

Wirth, M., Horn, H., Koenig, T., Razafimandimby, A., Stein, M., & Mueller, T. (2008).The early context effect reflects activity in the temporo-prefrontal semanticsystem. evidence from electrical neuroimaging of abstract and concrete wordreading. Neuroimage, 42, 423–436.

Wisniewski, L., Epstein, L. H., & Caggiula, A. R. (1992). Effect of food change onconsumption, hedonics, and salivation. Physiology and Behavior, 52, 21–26.